ESRA 2017 Programme
|ESRA Conference App|
Wednesday 19th July, 11:00 - 12:30 Room: F2 107
Analyzing the lives of LGBTI people - survey approaches to LGBTI persons, couples and families 2
|Chair||Dr Stephanie Steinmetz (Sociology Department, University of Amsterdam )|
|Coordinator 1||Ms Mirjam Fischer (Sociology Department, University of Amsterdam)|
|Coordinator 2||Ms Nancy Bates (Research and Methodology Directorate, U.S. Census Bureau)|
Session DetailsIn recent years, much progress has been made in the US, Europe and beyond with regard to legislation that is supportive and protective of LGBTs (Lesbian, Gay, Bisexual, Trans persons). While these achievements are laudable, it is important to keep evaluating to what extent they truly improve the lives of LGBTs. There are still many sources of inequality and discrimination that remain deeply embedded in the social fabric of societies. Compared to research on other minority groups, sexual orientation has been studied quantitatively much less in the social sciences. Yet, quantitative scholars have continuously made efforts to overcome the methodological challenges associated with studying this population by using surveys. This is an important development which should be encouraged and continued.
This session welcomes contributions focusing on the wide range of issues that need to be addressed when studying LGBT populations with survey-based methods. For example, is the 2-step method (sex assigned at birth and sex identify with now) best for reducing undercounting of transgender populations? As younger people embrace non-binary or gender-nonconforming identities, how must our questions and categories for gender identity change? Can sexual orientation be collected by proxy in surveys that use a single household informant to provide all member’s demographic information? How can concepts around sexual orientation and gender identity be translated for non-Western cultures and non-native languages? Are there interviewer effects when collecting SOGI in telephone and personal visit surveys? Does the addition of SOGI items harm unit response rates in surveys that do not typically collect such items (e.g. labor force or consumer expenditure surveys)? What are strategies for designing sampling frames intended at capturing LGBT populations? In addition, this session also invites submissions that focus on topical survey results around LGBT populations such as physical and mental health disparities, income inequality, hate crimes, and household and family structures. The session hopes to draw a cross-section of submissions from different countries and different survey experiences.
Paper Details1. Comparing two versions of a 2-step assessment for identifying transgender respondents in a national sample of U.S. adults
Dr Ilan Meyer (UCLA)
Mr Evan Krueger (UCLA)
Ms Stephanie Marken (Gallup)
Dr Sari Reisner (Harvard)
Dr Walter Bockting (Columbia University)
Dr Jody Herman (UCLA)
In recent years, transgender population surveys have gained interest from researchers and pollsters. A 2-step measure is most often recommended where respondents report their sex assigned at birth (male or female) and their current gender identity (man, woman, or transgender). Respondents are classified as transgender if their current gender identity is “transgender” or if their current gender identity is different than their reported sex assigned at birth.
We tested two versions of the 2-step question method that differ in the way the sex assigned at birth question is asked—version 1 leaving an open question and version 2 providing the “male” and “female” options in the question. The versions also differ in the response options regarding gender identity—version 2 including only “transgender” in addition to “man” and “woman”; version 1 including more response options (e.g., “genderqueer”). We compared these in a study of Gallup Daily Tracking survey, a nationally representative study of U.S. adults, aged 18 and over, conducted using a dual-frame, random-digit-dialing (RDD) sampling frame.
Version 2 was more efficient in reducing refusal and Don’t Know answers from 7% in version 1 to 0.1% in version 2. Version 1 yielded more respondents who identified as transgender than version 2 (0.53% vs. 0.37%). Compared to version 1, version 2 was not efficient at identifying non-binary/genderqueer individuals (0.4% vs. 0.06%). We also assessed whether the 2-step method of asking “sex assigned at birth” would be a good replacement for a standard Gallup assessment of sex. We found that against the standard sex question, both versions of the sex assigned at birth question performed extremely well, with sensitivity of over 99% for each.
2. Large-scale studies on small-scale groups: Sampling bias and other challenges in research on transgender individuals in the Netherlands
Dr Lisette Kuyper (The Netherlands Institute of Social Research)
While studies examining the wellbeing of lesbian, gay, and bisexual (LGB) participants are nowadays often based on population samples, studies of transgender individuals still mainly depend on community samples recruited from the offline and online transgender community. This sampling method introduces several biases, since transgender individuals involved in the community might differ from those who are not. Therefore, in recent years, several authors have argued to include questions on transgender background in representative population samples and have conducted field studies to assess the validity of questions such as the two-step method. The current paper aims to empirically address both the issue of sampling bias in community studies as well as the (im)possibilities of the use of general population surveys to map the wellbeing and living situation of transgender individuals in the Netherlands .
First of all, it will empirically examine the differences between transgender participants recruited from the transgender community (n = 209; 63.2% assigned male sex at birth; Mage = 40.13, SD = 13.59) and a sample of transgender participants recruited from a research panel (n = 62; 67.7% assigned male sex at birth; Mage = 42.86, SD = 13.88) in the Netherlands. Topics that will be addressed are sociodemographics, gender minority stress (such as level of openness about one’s transgender background, negative reactions received on one’s transgender identity, and transgender social networks), transgender background characteristics (such as possible medical transition steps or contact with professional care providers) and wellbeing (such as psychological distress and suicidality). This provides researchers on transgender issues insight into the biases in findings gained from community samples.
Secondly, the paper will map the opportunities to include items on transgender background in representative population studies in the Netherlands. For this, it is necessary to map the prevalence of transgender individuals in the Netherlands and to take in account the number of transgender participants needed to reliably address their experiences and living situation and the sample size of the population studies. While these issues are seldom a problem in large-scale countries like the U.S. (where most progress has been made on this issue), these factors do play a significant role in mapping the opportunities to use population samples in smaller-scale countries such as the Netherlands. The paper will show which Dutch population studies should be the primary focus to include items on transgender background to improve the data structure needed to reliably assess the wellbeing and living situation of transgender individuals in the Netherlands.
Limitations of the current data and implications for future studies aiming to provide reliable and valid data on the living situation and wellbeing of transgender individuals will be discussed.
3. Quality of Official Data on Cohabiting Same-Sex Couples in Germany
Dr Andrea Lengerer (GESIS - Leibniz Institute for the Social Sciences)
Same-sex couples are a small group, both in absolute terms and relative to the total population. A representative study of same-sex couples requires data based on a very large sample size, which is typically the case with official data. However, it is questionable whether same-sex couple relationships are reported reliably in official surveys. Despite the increasing social and legal acceptance of these relationships, they remain a sensitive topic and are thus likely to be underreported. Moreover, same-sex couple relationships are often asked about only indirectly in official surveys, which mean that the respondents do not know exactly whether they should report them at all.
Against this background, the presentation focuses on the quality of official data on cohabiting same-sex couples. It is hypothesized that same-sex cohabitation is underreported in official surveys. Furthermore, it is expected that the extent of underreporting decreases over time, as tolerance and acceptance of same-sex living arrangements increase. Data from the German microcensus are used to test these hypotheses empirically. The findings are transferable to official data of other countries where cohabiting same-sex couples are measured in a similar way.
A comparison of microcensus data with external data is made for registered same-sex couples. However, no reference data are available for all other (i.e. non-registered) same-sex couples. Hence, the living arrangement of the respondents is estimated by using information about the composition of the household. It is checked whether there is a possible partner of the same or opposite sex within the household or not. Based on this, the responses to the (voluntary) question about a partner in the household are analyzed.
As expected it was found that respondents with a possible partner of the same sex in their household were less willing to report this partner in the microcensus. This was particularly the case in face-to-face interviewing, which is the predominant survey mode in the microcensus, and it indicates underreporting of cohabiting same-sex couples. However, little evidence was found of selective reporting of these couples. Underreporting of cohabiting same-sex couples was found to have decreased only slightly over time, so that the increase in this living arrangement that has been observed in recent years cannot be attributed solely to an increase in the willingness to report.
The work is part of a research project on the prevalence, trends and social conditions of same-sex partnering in Germany, which is funded by the German Research Foundation (DFG).
4. Sensitivity and specificity of a 1-item assessment of LGBT identity in a national sample of U.S. adults
Ms Stephanie Marken (Gallup)
Mr Evan Krueger (UCLA)
Dr Ilan Meyer (UCLA)
Dr Walter Bockting (Columbia University)
Dr Jody Herman (UCLA)
Dr Sari Reisner (Harvard)
A vital first step in measuring issues of importance to the lesbian, gay, bisexual and transgender (LGBT) population is to accurately identify individuals’ gender identities and sexual orientations. Survey researchers have experimented with several methods of identification to determine which approach best covers the LGBT population while minimizing respondent burden.
We tested two approaches: a one-question assessment of LGBT identity versus 3 questions to measure sexual orientation and gender identity. In the one-item approach, respondents are asked if they identify as lesbian, gay, bisexual or transgender, providing a “yes” or “no” response. The 3-items approach assessed sexual orientation and gender identity separately. To assess sexual orientation respondents were asked “which of the following do you consider yourself to be?” Respondents select from multiple response options, including “straight or heterosexual,” “lesbian,” “gay,” “bisexual,” “queer,” and “same-gender loving.” To assess gender identity, respondents were first asked what sex they were assigned on their original birth certificate. Then, they were asked to choose the label that describes their current gender identity, with the response categories, “woman,” “man,” “transwoman,” “transman,” and “non-binary/gender queer.” The one-item approach has obvious economic advantages as it provides more efficient data collection. However, this approach may not be as precise and may lead to undercoverage of the population.
To measure the sensitivity and specificity of the one-step assessment, Gallup worked with researchers at UCLA and other universities to compare the two approaches using the Gallup Daily Tracking survey, a nationally representative study of all U.S. adults, aged 18 and over, conducted using a dual-frame, random-digit-dialing (RDD) sampling frame. We present differences between the two approaches based on data from approximately 7,700 respondents. We will present findings from this research, including interviewers’ experiences implementing each approach and the accuracy associated with covering the population using the different methods.
5. The proportion of the population of England self-identifying as lesbian, gay and bisexual in England: producing modelled estimates based on national social surveys
Dr Sanne Christine van Kampen (University of Plymouth)
Mr Mauro Fornasiero (University of Plymouth)
Dr William Lee (University of Plymouth)
Dr Kerryn Husk (University of Plymouth)
There is now a body of published research on health, inequality and discrimination among the lesbian, gay and bisexual (LGB) population. However, reliable data on sexual identity and sexual orientation is absent in England. In order to assess health outcomes by sexual orientation, there is a need for a widely accepted national estimate of the size of the LGB population. In recent years, progress has been made in developing a general approach to quantify the LGB population and an increasing number of national surveys now include this measure. Still, recent national estimates vary widely from 1.9% provided by the Integrated Household Survey (ONS, 2015) to 9% as reported by an online opinion survey (YouGov, 2015). Even after standardisation of question wording, variety in the estimates is to be expected due to methodological differences in the selected survey population, sampling methodology, sample size, and eligibility of the participants to answer the question. Besides, non-response and misreporting are considered to be the central issues regarding the quality of sexual identity responses. In this study we adapted a process created for other high-need groups (minority ethnic groups) and used available data to estimate as accurately as possible the prevalence of LGB in the population of England. This allowed for more accurate health need prediction and, for the first time, a synthesised national estimate of the total size of the LGB population in England.
This study systematically identified and pooled nationally representative surveys which included a question on sexual orientation. Twenty two surveys were identified of which 15 were suitable for pooling as they were representative of the general population of England. An aggregated mean estimate was produced using weights based on sample size, response rate and proportion of missing data. The national weighted estimate of individuals identifying as LGB or ‘other’ was 2.50% with a theoretical upper limit of 5.89% (if all non-responders are assumed to be LGB). This translates to an estimated 1,36 million people self-identifying as LGB or 'other' in mid-2014 in England. The proportion was highest in men (3.10%), young adults up to 34 years (4.74%), mixed/multiple (4.25%) or other ethnic groups (5.50%), and the London region (4.26%). The reported 2.50% is likely to be an underestimate as individuals may choose to misreport their sexual identity in survey and monitoring settings influenced by perceptions of confidentiality and social acceptance. Future studies should assess the effects of survey context and mode of administration.This report specifically focused on the LGB population and further work is needed to include transgender and intersex people. The results of this study could inform public health institutes, government offices, advocates and academic groups, and help to improve monitoring of health and social inequalities among the LGB population.